Short answer

On upload, Civic.ly reads the photo's GPS metadata, drops the asset at those coordinates, runs AI auto-classify to propose a name and description, then enriches the location: if the GPS point falls inside an area you've drawn on the map, the asset inherits that area's location tag; otherwise it tags the nearest street; in both cases it also gets a what3words address. GPS is approximate, so expect to nudge each pin to its real location on the web app afterwards.

Detail

The end-to-end placement flow on upload:

  1. Read GPS from the photo. Civic.ly reads EXIF metadata. If the photo doesn't have GPS, the asset comes in without a location and has to be placed manually.
  2. Drop the asset at the GPS point. This is the starting position — usually close, sometimes off by 5–20 metres in built-up areas.
  3. Run AI auto-classify on the photo. The AI proposes the asset type (Wheelie Bin, Picnic Table, Defibrillator Cabinet, etc.), a description, and a first-pass condition assessment.
  4. Apply area-based location tag. If the GPS point falls inside a polygon you've drawn (e.g. Memorial Park, Jubilee Park), the asset inherits the area's location tag automatically. This is a powerful reason to draw your areas first — every asset added inside one is auto-tagged.
  5. Otherwise, apply a street-based location tag. If the asset isn't inside any drawn area, Civic.ly tags it to the nearest street (e.g. New Road, Munsley Road).
  6. Apply a what3words address as a precise locator that's easy to share by phone or radio.

The result: from a single photo, you get an asset on the map, named and described, area-tagged, street-tagged, and three-word-addressed. Then you nudge the pin if needed.

Watch-outs

  • GPS accuracy is variable. Cloud cover, dense buildings, tree cover, indoor capture, or simply standing back from the asset to fit it in frame all introduce drift. Always do a pin-nudge pass after upload — much faster than trying to capture perfect locations in the field. Full process: After uploading assets from a phone, do I need to adjust their map locations?.
  • Areas have to exist before assets land in them. If you upload bins and then draw the park boundary, the bins won't be retroactively tagged. Best practice: draw your areas first (parks, depots, allotments, churchyards), make sure each area's Location tag field is set, then start uploading. See How do map areas work, and how do location tags get inherited?.
  • AI classification can run on upload but may show as NOT_CLASSIFIED for a while. It's an asynchronous process — give it a moment. If it stays not-classified for many minutes, something probably went wrong (no photo, weird format).
  • Asset names from AI are starting points, not gospel. Always do a sanity-pass after a batch upload — particularly to add operational codes (CS numbers, asset IDs) into the name where the council uses them.